ML20212K664

From kanterella
Jump to navigation Jump to search
Inadvertent Recirculation Actuation Signals at C-E Plants, Technical Review Rept
ML20212K664
Person / Time
Site: Calvert Cliffs, Arkansas Nuclear, 05000000
Issue date: 08/07/1986
From: Cintula T
NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD)
To:
Shared Package
ML20212K662 List:
References
TASK-AE, TASK-T606 AEOD-T606, NUDOCS 8608220336
Download: ML20212K664 (8)


Text

r ~

AE0D TECHNICAL REVIEW REPORT 1/

UNITS: Arkansas-2 and Calvert Cliffs-2 TR REPORT NO.: AE00/T606 DOCKET N05.: 50-368 and 50-318 DATE: August 7, 1986 LICENSEES: Arkansas Power & Light Co. EVALUATOR / CONTACT: T. Cintula Baltimore Gas and Electric Co.

NSSS/AE: Combustion Engineering /Bechtel

SUBJECT:

INADVERTENT RECIRCULATION ACTUATION SIGNALS AT COMBUSTION ENGINEERING PLANTS EVENT DATES: January 1 and May 23, 1985

SUMMARY

The cause and potential safety implications of separate events involving an inadvertent recirculation actuation signal (RAS) at two Combustion Engineering designed plants are discussed. At one plant, without antidraindown check valves in the emergency core cooling system pump suction lines, a significant inventory of borated water in the reactor water tank (FWT) was transferred to the containment sump by the inadvertent RAS. Antidraindown check valves prevented a similar loss of RWT inventory at the other unit. A review of each event, including a postulated subsequent safety injection actuation signal (SIAS), did not identify a safety concern for an inadvertent RAS event for either plant design.

l INTRODUCTION On January 1,1985, a spurious actuation of the recirculation actuation system at Arkansas Nuclear One, Unit 2 (AN0-2) resulted in the draining of approxi-mately 50,000 gallons of borated water from the RWT to the containment sump.

On May 23, 1985, a similar spurious actuation at Calvert Cliffs Unit 2 did not result in a transfer of bcrated water from the RWT to the containment surp. A review of the respective Combustion Engineering (CE) plant designs revealed that a draindown at the Calvert Cliffs plant did not occur because antidraindown check valves are installed in the containment sump suction lines while a .

draindown did occur at ANO-2 because antidraindown check valves are not installed in the containment sump suction lines. In view of the design dif-ferences, a review was initiated to evaluate the potential safety implications of an inadvertent PAS actuation at both plants. This safety review evaluates the potential consequences of an inadvertent RAS during a postulated loss of coolant accident. The purpose of this assessment was to evaluate whether an inadvertent RAS during a postulated accident has the potential for causing common mode failure of the engineered safety features pumps due to inadequate net positive suction head (NPSP).

1/ This document supports ongoing AE00 and NRC activities and does not repre-sent the: position or requirements of the responsible NRC program office.

8605220336 G60007 ' ~

{DR ADOCK 0300031g PDR

DISCUSSION

1. Arkansas-2 On January 1, 1985, with AN0-2 at 100 percent power, a spurious RAS caused approximately 50,000 gallons of borated water to inadvertently drain by gravity flow from the RWT to the containment sump. The event occured during a routine monthly surveillance test of the engineered safety features actuation system (ESFAS). The spurious RAS was initiated when an instrument technician de-energized an RAS logic matrix relay in accordance with approved procedures.

This caused an intented " half-leg" actuation of the ESFAS RAS. However, at this time, a coincident RAS logic matrix relay also de-energized. This satisfied the "2-out-of-4" RAS actuation logic which opened the containment sump suction valves and closed the RWT suction valves in the redundant emergency core cooling system (ECCS) pump suction lines. This caused the l *ransfer of borated water from the RWT to the containment sump. The draindown i

of water continued while both sats of valves were cycling and was terminated when the RWT valves reached the 'ully closed position. During the event, the two large diameter (i.e., approxuately 24") RWT outlet lines allowed a significant quantity of borated water to gravity drain from the RWT to the I containment sump during RAS valve realignment.

l An RAS-initiated valve realignment begins when each of the motor-operated containment sump suction valves (see Figure 1) receives a signal to open.

These valves require approximately 22 seconds to fully open from the normally closed position. The motor-operated RWT outlet valves do not begin to close i until they receive a signal that the containment sump valves are fully open.

! This valve sequence assures that adequate NPSH is maintained for the operating a ESFAS pumps. Once the containment sump valves are fully open, the RWT outlet H valves are signalled to close and require approximately 80 seconds to stroke completely closed. Therefore, for 102 seconds following the RAS, both the containment sump valves and the RWT cutlet valves are at least partially open.

While both sets of valves were partially open during the event, approximately 50,000 gallor.s of borated water drained from the higher elevation RWT (located ,

at grade level outside of the containment building) to the lower elevation containment sump.

To prevent recurrence, a procedural change was implemented rcouiring that the RWT outlet valves be manually closed and locked closed before and during

! testing of the RWT level switches. With the RWT outlet valves locked closed, an inadvertent RAS during relay testing would not result in borated water draining from the RWT to the containment sump.

2. Calvert Cliffs-?

On May 23, 1985 with Calvert Cliffs Nuclear Power Plant Unit 2 at 100 percent power, an instrument technician inadvertently initiated an PAS during surveillance

testing of the RWT level switches. The first RWT level switch which was tested appeared to be functioning incorrectly and was left in the tripped condition.

The technician proceeded to remove the cover from a second switch to allow it to be tested. The second switch tripped during removal of its cover. Because the first RWT level switch had been left in the tripped condition, the actuation of the second switch satisfied the "2-out-of-4" RAS actuation logic signalling the containment sump valves to open. When the sump valves were fully open, the.

RWT outlet valves were signalled to close. However, during the time that both valves were open, no borated water was transferred from the RWT to the contain-ment sump. At Calvert Cliffs, antidraindown (reverse flow) check valves prevented the borated water from flowing into the containment sump after the containment sump valves opened. The antidraindown' check valves are located in the suction headers between the PWT and containment sump upstream of the normally closed isolation valves for the containment sump.

ANALYSIS AND EVALUATION Purpose and Operation of the Recirculation Actuation Sianal The purpose of the RAS at CE plants is to automatically transfer suction of the operating high pressure safety injection (HPSI), and containment spray (CS), pumps from the RWT to the containment sump and, to trip the low pressure safety injection (LPSI) pumps. The RAS actuation occurs on a sensed low RWT liquid level. An RAS terminates the injection phase of the engineered safety features (ESF) pumps while they still have adequate NPSH from the remaining inventory in the PWT. The RAS changes the valve alignment to the post-accident, long-term cooling mode, to enable the HPSI (and possibly,'the CS) pumps to take suction from prev W ' injected emergency core coolant th t would have spilled out the brc;r, ano u .lected in the containment sump. As shown in Figure 1, a typical CE plant design incorporates two independent ECCS pump suction lines.

Both lines will switch over from the PWT to the containment sump on an RAS. The RAS actuation logic incorporates two separate instrument channels. Each channel is provided with two RWT level switches (and associated logic relays). Four level switches are provided in all. The RAS logic requires any 2-out-of-4 switches (relays) in the two channels to change state to cause an RAS.

~

The kWT n"tlet valves are normally open to allow borated w'ater from the PET to fill the ECCS suction piping up to the seat of the containment cump valves or (deperding on plant design) the antidraindown check valves. The RWT also maintains the pump flow path full up to the normally closed injection or spray valves. The valve realignment begins with the containment sump suction valves "

rec iving a signal to open. At ANO-2, a typical CE plant, the 24" motor-operateu m suction valves require about 22 seconds to fully open from their -

normally closed paition. When the containment sump valves begin to open at a plant without antidrainn wn -%.k valves (e.g., ANO-2), barated water will begin to drain from the RWT to the containment sump. The motor-operated RWT outlet valves do not begin to close until they receive a signal that the containment 3 ump valves have fully opened. This valve sequence ensures that adequate NPSH is maintained for the operating ESF pumps. The 20" diameter RWT outlet valves need an additional 80 seconds to completely close. Therefore, from the time an Rf 5 occurs, approximately 102 seconds will elapse in which the RWT outlet valves will be either fully or at least partially open. During this 102 second period, at plants without antidraindown check valves, a significant quantity of borated water can drain by gravity from the elevated RWT (located at grade level outside of the containment building) to the containment sump. However, the large water volume transferred does not have the potential for flooding vital equipment. The containment is designed to accommodate the combined water volumes resulting from the depletion of the RWT, injection from the safety injection tanks (SITS) and the water which would be lost from the reactor vessel during a postulated LOCA.

O

L . _

Effects of an Inadvertent Actuation on Makeup Inventory At ANO-2, the 50,000 gallons of borated water which drained from the RWT caused the PWT level to decrease fron 98 to 88 percent of full capacity levels. The RWT has a maximum volume of 500,500 gallons of which 410,000 gallons are considered as available for safety injection and containment spray. This capacity is sufficient for the ESF pumps to take suction from the RWT for a minimum of 30 minutes after an SIAS with all of the ESF pumps operating at rated flow. A 10 percent reduction in RWT inventory, caused by an invertent RAS actuation, would reduce the injection phase of a design basis accident by approximately 3 minutes. However, the borated water from the inadvertent RAS would be available in the containment sump for the recirculation mode if needed. Therefore, the net safety effect of an inadvertent RAS at a plant without antidraindown check valves during a design basis accident would " a shortened injection phase with the same total volume of borated water available to sustain long-term core cooling.

Clearly for plants with antidraindown check valves, there is no RWT level reduction potential for a spurious PAS actuation. At Calvert Cliffs, for example, no water would be transferred to the containment sump. The borated water in the RWT would remain in place, seating against the antidraindown check valves. The containment sump valves would cycle open and the RWT outlet valves would subsequently close without transferring RWT inventory. Finally, follow-ing an inadvertent RAS, several status indicators would alert the control room operators of the need to reopen the RWT outlet valves.

Potential for Line Voiding and ESF Pump Gas Binding The potential for an inadvertent RAS actuation coincident with an ESFAS actua-tion (on an accident signal) leading to vapor binding of the ESF pumps at a plant with antidraindown check valves was also assessed. This scenario was not reviewed for plant designs without antidraindown check valves in the ECCS suction lines because, under any postulated sequence of RAS /SIAS signals, adequate inventory of borated water would be available in either the RWT or containment sump to assure adequate ESF pump performance.

  • The postulated scenario for plants with antidraindown check valves involves an SIAS. occurring shortly before or after an inadvertent PAS. In this situation the ESF pumps may be running while the RWT outlet valves are slowly closing following the inadvertent RAS. The basis for the concern associated with this scenario is that after the motor-operated RWT outlet valves receive a signal to close (from the inadvertent PAS) the valves will not be capable of receiving an opening signal (from the SIAS) until the valves have travelled to their fully closed positions. With restricted replenishment flow from the PWT passing through the closing RWT outlet valves, the operating pumps would attempt to deplete the water trapped in the ECCS suction lines. The inventory removed from the suction lines would be discharged via either the combined injection flow into the cold legs, or the recirculation flow back to the RWT. Depending on the initiating cause of the SIAS, the concern would be that sufficient water might be removed to cause pump vapor binding before the RWT outlet valves .

reopen sufficiently to refill the ECCS suction lines via flow from the RWT.

Without sufficient pump flow and inventory replacement in the lines, it might -

be possible for pump heat and/or lack of pump flow to cause a steam void in

_ the pump casing. This in turn might cause the operating pumps to lose adequate b%--_ _ _ _ _ . _ , - - - _ _ . . _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ . _ _ - _ - _ _ . _ . - _ _ _ _ _ _ _ _ _ _ . _ _ _ _ . _ _ _ _ _ _ . _ _ _ _ - _ _ _ . _ _ _ _ _ _ - _ . _ _ _ m _ __. _ _ _ _ _ _ _ . _ _ - . _ _ _ _ - - -

suction and become vapor bound. In addition, extended pump operation in a vapor bound. condition may cause common mode pump damage or failure even without permanent pump damage or contribute to subsequent long term failure. The disruption of injection flow in the early mitigation phase of a LOCA due to J pump vapor binding might lead to core damage in excess of that previously calculated.

To evaluate the potential for sufficiently depleting the volume of borated water trapped in the ECCS suction line and subsequently steam vapor binding all of the operating pumps, the action of each automatic signal and maximum pump performance for a typical CE plant were first reviewed. The actions involved in an RAS and SIAS actuation are as follows:

Recirculation Actuation Signal (RAS)

1. LPSI pumps trip
2. HPSI miniflow valve to RWT shuts
3. HPSI pumps transfer suction from RWT to containment sump Safety Injection Actuation Signal (SIAS)
1. HPSI pumps start and their niniflow recirculation valves to the RWT open
2. LPSI pumps start and their miniflow recirculation valves to the RWT open
3. HPSI and LPSI (after permissive) injection valves open 4 PWT outlet valves receive a confirmatory open signal In addition, the following valve repositioning would take place for a combined RAS /SIAS actuation:
1. Containment sump valves open
2. RWT outlet valves close
3. RWT outlet valves reopen To conservatively calculate the inventory removed from the ECCS pump suction lines, the following performance data were used:

Engineered Safety Features Pump Performance Data The highest absolute pump flow rate at either ANO-2 or Calvert Cliffs is as follows:

LPSI pump miniflow rate: 2 pumps x 100 ppm / pump = 200 gpm LPSI pump maximum injection flow rate: 2 pumps x 5200 gpm/ pump = 10,400 gpm HPSI pump miniflow rate: 3 pumps x 30 gpm/ pump = 90 gpn

~

HPSI pump maximum injection flow rate: 3 pumps x 825 gpm/ pump = 2475 gpm

, , - -e

" 6@ipF

. (, .

Minimum Trapped Volume of ECCS Pump Suction ~ Line 43' of containment piping + 38' elevation. difference for each of two 24" diameter suction lines = 3350 gallons Two bounding scenarios were considered for depleting the ECCS suction lines:

(1) all ECCS pump miniflow recirculation to the RWT with the injection check valves closed (no injection flow); and (2) maximum pump injection into the reactor cold legs (no minimum flow to the RWT). In the first scenario, the most conservative combination of the largest miniflow recirculation of all ECCS pumps with the shortest possible length of ECCS suction piping would permit at least 12 minutes of miniflow recirculation. This is a sufficiently long time to assure that the RWT outlet valves would reopen before the ECCS suction line is sufficiently depleted to cause pump vapor binding. Therefore, pump NPSH would be maintained during miniflow recirculation conditions independent of the cycling of the RWT outlet valves and the pumps would not be subject to steam binding.

In the second scenario, conservative calculations assuming no replenishment from the RWT, [using the maximum injection flow rate for all operating (HPSI &

LPSI) pumps], with shortest possible length of ECCS suction piping show that the trapped water volume could be evacuated in about 15 seconds. This very conser-vatively calculated time period is too brief to assure that the RWT outlet valves would not be in the closed position or in the process of closing while the pumps are taking suction. However, although the RWT outlet valves may not be sufficiently open to supply adequate inventory to the ECCS supply header at maximum ESF pump flows, further calculations indicate that the large RWT outlet butterflyvalvescansupplytheequivalentofthecombinedminiflowrgte 4'

requirements for all of the pumps after opening only a few degrees (4 ) from the fully closed position. As long as individual pump miniflow requirements-are satisfied in the injection phase, it should not be expected that pump vapor binding or pump damage would occur. Thus, it would appear that the combination of the large volume inventory of the ECCS suction lines and the short time duration (after the RAS logic is completed) that is needed .for _ the RWT outlet valves to recycle to a position where pump minimum flow requirements are satis-fied for each pump, leads to the conclusion that pump vapor binding would not occur in these postulated bounding scenarios.

FINDIflGS AND C0f!CLUSIONS RWT designs without antidraindown check valves can result in the draining of significant quantities of borated water from the RWT to the containment sump during an inadvertent RAS. The large volume of water is due to the: (1)large inner diameter of the piping from the RWT to the containment sump; (2) sequential cycling of the containment sump valves and RWT outlet valves; and (3) long cycle duration of the large diameter suction valves. However, the large volume of water transferred to the containment sump does not present a potential for reducing the total inventory of borated water for long-term core cooling or for flooding vital equipment within the containment enclosure. Similarly, an inadvertent RAS during normal operation at a plant without.antidraindown check valves could result in the need to pump out the containment sump, processing the water to RWT standards and returning the water.to the RWT.- Actions taken at ANO-2 will reduce the likelihood of an inadvertent draindown of the RWT from recurring. CE plants with antidraindown check valves i_n the ECCS suction lines prevent inventory from being drained from the RWT on an inadvertent RAS and there were no apparent consequences to an inadvertent RAS at these plants.

. . _ . _ .= .-

7-An investigation of various accident sequences that might be adversely affected by a postulated inadvertent RAS (either before or af ter an SIAS) did not identify any new safety concerns related to ECCS performance for CE plant designs with or without antidraindown check valves.

SUGGESTION The procedure change implemented at ANO-2 namely, to require closure and locking of the RWT outlet valves before and during testing of the RWT level switches, would appear to be an effective means of preventing an inadvertent RWT draindown during the required periodic testing of these sensors. Licensees of pressurized water reactors (PWRs) without antidraindown check valves may wish to consider implementing a similar preventive measure during. testing of their RWT level switches.

It is also suggested that consideration be given to including a description of the events at ANO-2 and Calvert Cliffs and the measures taken at ANO-2 to prevent recurrence in a future edition of Power Reactor Events.

8 9

e AWT 6 sautoown i >

COOLING MtATSICHAMott '

0

' ct 4 1r j '/ X nYs!CLcst CPEN dk pmCM F5 OI#AI AT e g e

-h-CPIN

-- gi.

=-

clat

.SP -

to v e

,g g y stant LPSI MSAT SICHANoth

-k

.n . . . = ......,,,,

I X

9IA8 h 1r ""'

= '+ 2 _

C IN v x 8 01AlstAR, ,

e '#

h eats e

A

. A.

C9tN'

.+ k

- ^

mi _2 x .,A......,,

. A.. 1

,N M U" LPsi g

it.o i ,,,

/:["

s't,.' x m e..,An, M ,

6 , ErrutR noron.

mas.stop OPERATE 3 VALVES Si .5i Antr $rnomm CHECK VALy[$

1 evue 3A's CPtn

[

T FIGURE 1. TYPICAL CE ECCS-CSAS PUMP SUCTION PIPING s.

- - -- --